Few or No Results — What to Do?

You've started a search and you're getting fewer leads than expected — or a run that ends without any results? This is almost always fixable. In most cases it's not "there aren't enough companies out there" — it's that Leadscraper doesn't yet know exactly what makes a good lead for you. This article shows you what's behind that and how to fix it step by step.

First, Understand How Leadscraper Searches

Leadscraper isn't a database dump where ready-made lists are served up. Every search goes through the following:

1. Think first: Leadscraper plans its own search strategy per request — which industry directories, association registers, and web sources make sense and why. You can watch this reasoning live in the activity stream ("Goal", "Strategy", "Reasoning", "Curated Source").
2. Then search: The system searches the open web, German industry directories, association member lists, and company websites — live, fresh for every run.
3. Then evaluate: Every single candidate is checked against your request. Only what truly fits is delivered.

This explains two things: first, a run deliberately delivers up to 10 checked leads rather than hundreds of unfiltered addresses. Second, Leadscraper gets more precise with every rating you give. Few results are therefore usually a signal that something needs adjusting in the request or in the training — not the end of the road.

The Three Levers, in the Right Order

If a run comes back thin, work through them in this order: start broader → rate → get more specific. That's exactly how the system is built.

Lever 1: Start Broader

The most common reason for empty or thin runs is a request that's too narrow or too specifically phrased. Multiple filters at once (very narrow industry + small city + specific headcount + specific role) cut down the search space so much that barely any candidates remain.

How to make the request broader:

• Everyday language rather than jargon. Describe the company the way a customer would name it. "Packaging manufacturer" finds more than a very narrow niche technical term. You can always narrow it down later.
• Expand the region. Instead of a single city, try a federal state or the entire DACH region (Germany, Austria, Switzerland). A small city alone simply may not have enough matching companies.
• Drop optional filters. Leave out headcount or a very specific role for now. You can teach these criteria to the system later via feedback — more on that in a moment.
• A search isn't a filter form. You don't have to narrow everything down at once. Better to start broad and sharpen Leadscraper over multiple runs.

Tip: When you submit a request, Leadscraper briefly checks it for searchability upfront. If it's structurally too broad (classic: "IT service providers Germany") or too vague, you get a notice before the run with concrete suggestions for how to sharpen it. Take these suggestions seriously — they save you an empty run.

Incidentally, Leadscraper already tries to rescue thin searches on its own: if the strategy finds too few candidates, the system automatically adds more search queries and pulls in related industries before giving up. If a run still comes back empty, the system has honestly decided that the request in its current form isn't workable — and typically suggests adjusted phrasings.

Lever 2: Rate — and With a Reason

This is the most important and most underestimated lever.

Never judge quality based on a single run. The first run is a cold start: Leadscraper doesn't know your preferences yet and delivers based on your plain request. Your feedback is what makes the system good.

How to do it:

• Rate every lead. Accepting is implicit — moving a lead further through the pipeline signals it fits. Reject non-matching leads with "Not a fit".
• For "Not a fit", always give the reason. Choose the matching reason (e.g., wrong industry, wrong role, wrong region, wrong size, not an operating company, no buying signal, already known, poor data) or write a short comment. The reason is key: that's what the system learns why something doesn't fit — not just that it doesn't fit. Ratings can be reset at any time.
• Moving leads forward is training too. Every lead you process further tells the system what a good match looks like.

What happens in the background: your ratings are collected and condensed into patterns (acceptance and rejection patterns by industry, size, region, role, business model, and more). These patterns flow into the next run and guide the evaluation of every candidate.

The learning curve in numbers:

In practice: an accuracy rate that sits around 60–70% at cold start rises toward 85–90% with around 25 ratings, and beyond. The first 10 ratings are the most important investment — from there, the learning kicks in. Anyone who stops rating after the first run is skipping exactly the mechanism that makes Leadscraper better than a static list.

Lever 3: Get More Specific

Once you've rated a few runs, you can deliberately sharpen things without starting from scratch:

• Refine the request per run. You can narrow a search for a single run (e.g., "folding carton manufacturers NRW, purchaser") without permanently changing your base profile. Leadscraper recalculates the request at runtime.
• Build on learned patterns. What you've taught via feedback is retained and keeps working. You don't have to write criteria like "no agencies" or "only family-run businesses" into every request if you've rated them consistently — the system carries them forward.
• Add your own company. If you provide your website URL in the setup, Leadscraper better understands your offer and can give more fitting explanations for why a lead is a match for you.

Understanding "Soft Keep": Yellow-Flagged Leads

Some leads get a yellow note (caveat). These are results that probably fit but aren't 100% confirmed — for example, because information on the company page wasn't unambiguous. They're only delivered if there are still spots available after the definite matches. Treat them as candidates to check: rating them helps the system especially in sharpening its grey zones.

Quick Checklist for Thin Runs

1. Is the request too narrow? Phrase it more broadly (everyday language, larger region, fewer filters).
2. Did you pay attention to the upfront notice when submitting, if one appeared?
3. Did you rate all leads in the run — including the rejection reason?
4. Have you done multiple runs? Only after a few ratings does the judgment become reliable.
5. Is personalization already active (from 10 ratings)? You can see the status in the maturity indicator in the Memory section.

Frequently Asked Questions

Why do I get up to 10 leads per run and not more?
Leadscraper deliberately prioritizes checked quality over volume. Every run delivers up to 10 leads, each individually evaluated against your request. More volume comes through more runs — and those get more accurate with your feedback run by run.

My first run was disappointing. Is that normal?
Yes. The first run is the cold start with zero personalization. Don't judge quality until you've done multiple runs plus ratings — that's not a consolation prize, it's how the system works.

I'm not getting the same companies as last time — why?
Leadscraper remembers already-delivered companies per profile and doesn't deliver them again in subsequent runs. This way you get new leads with repeated runs rather than duplicates.

Does it help to phrase the same search slightly differently a few times?
Yes. Different phrasings open up different sources and candidates. Combined with consistent rating, this is one of the most effective ways to get more and better matches.

Which countries does Leadscraper cover?
The DACH region: Germany, Austria, and Switzerland. If your target region is outside that, that explains the missing results.

How can I see whether the system has already learned something?
In the Memory section, the maturity indicator shows you how many ratings you've given and which patterns are active. In the activity stream of a run, you can also see whether personalization is working ("X ratings · Y acceptance patterns · Z rejection patterns active in reasoning").

In short: few results are rarely an endpoint. Start broad, rate consistently with a reason, and then sharpen from there. Leadscraper is designed to get better with every rating you give — use that.

Related articles: "How Does the Training / Personalization Work?", "How Do I Phrase a Good Search Request?", "Rating and Accepting Leads".

Updated on: 23/06/2026